Indirect interactions are a direct result of herbivore-induced plant responses which provide the mechanism to mediate a range of interactions between the inducing species and species that are subsequently affected by the plant responses. Here we will remain focused on insect herbivore-induced plant responses, although it is worth noting that plants may be induced by other
stressors, including mammalian herbivores, pathogens and mechanical damage.
Likewise, diverse organisms may be at the receiving end of the interactions, including pollinators, pathogens and non-insect herbivores.
Plant-mediated indirect interactions between insect herbivores
Plants mediate interactions ranging from mutualisms to competition, although the focus of research has been on the latter. In 1960, Hairston et al.
published a seminal theoretical paper that altered the conventional view of the factors limiting herbivore abundance. Prior to that time, herbivore abundance was assumed to be limited primarily by resource availability (bottom-up hypothesis), so that competition with other herbivores for resources was considered the most important interaction. However, Hairston et al. (1960) argued that herbivore abundance is unlikely to be limited by resource competition because plants are clearly abundant and rarely consumed entirely. They proposed instead that herbivore abundance was dictated by interactions with higher trophic levels, such as predation and parasitism.
This idea that top-down forces control herbivore populations, known as the
“Green World” hypothesis, persisted for decades. This view has slowly shifted as understanding of plant trait changes induced by herbivory has grown. Induced resistance provides a mechanism for competition between herbivores that does not depend solely on food quantity but rather on the herbivore-induced changes in food quality. In addition, ecologists are beginning to recognize that positive indirect interactions are also quite common and may be equally important in structuring ecological communities. Induced susceptibility provides a mechanism for positive indirect interactions between insect herbivores, as mediated by the plant food source.
Plant-mediated indirect interactions do not necessarily depend on two insect species utilizing the same part of the plant or sharing the same plant simultaneously. As induced responses can occur at time scales ranging from hours to years, the length of temporal separation can be highly variable. Altered plant chemistry generally occurs quite quickly and may relax to constitutive levels within days or weeks. Morphological changes including increased trichome
densities or new growth may occur quite slowly in comparison but they generally persist for a much longer time span.
Interactions between temporally separated herbivores have been reported since at least 1986 when Faeth documented asymmetric competition between early and late season herbivores of Quercus emoryi. He found that early-season leaf chewers negatively impacted late-season leaf miners. This impact occurred more frequently on intact leaves rather than damaged leaves because damaged
leaves had higher tannin and lower protein concentrations compared to intact leaves. The observed interaction between the two herbivores was strongly asymmetric, approaching amensalism.
Since publication of Faeth’s 1986 paper, many other researchers have documented interactions between temporally separated herbivores as mediated by plants. There are examples of both negative and positive interactions
involving a variety of herbivores including aphids, beetles, planthoppers, weevils, sawflies, galling insects and thrips (Ohgushi 2005). For instance, a stem-boring moth (Archinara geminipuncta) induced narrow side shoots in common reed (Phragmites australis) which were then utilized by the gall-making midge
Giraudiella inclusa which prefers thin shoots. Prior moth damage was correlated with increased midge egg survival and abundance, demonstrating induced susceptibility (Tscharntke 1988).
More recent work has focused on how sequence of arrival modifies interactions. A few studies (Voelkel and Baldwin 2004, Viswanathan et al. 2007) have indicated that sequence of arrival may be an important factor in indirect interactions, however few have tested this explicitly. A recent meta-analysis of aboveground–belowground interactions found that the sequence of arrival strongly influenced interactions between aboveground (AG) and belowground (BG) herbivores. BG herbivores had a positive effect on AG herbivore
performance when they arrived simultaneously but not when either arrived first.
AG herbivores on the other hand had a negative effect on BG herbivore
performance when AG herbivores arrived first (Johnson 2012). In a study of the
interaction between leaf- and root-feeders on teosinte and cultivated maize, the leaf herbivore only negatively impacted colonization and growth by the root-feeder when the leaf herbivore arrived first. When the root-root-feeder arrived first, larval performance was not affected although adult emergence was reduced. The authors proposed that feeding deterrent or repellent secondary metabolites are the cause (Erb et al. 2011). In general, the competitive advantage goes to the earliest colonizing herbivores.
Plant mediated interactions frequently occur between insect herbivores that do not utilize the same plant part. Like temporal separation, spatial
separation between interacting species is highly variable as induced plant responses may be localized to the damaged tissue or be systemic. The scale of variability in plant tissue quality ranges from within a single leaf to the entirety of the plant body (Shelton 2005). Spatially separated herbivores may even be mediated by different plants, with plant-plant signaling inducing defense
responses across individuals. Spatial separation also encompasses interactions between members of different feeding guilds that frequently utilize different parts of the host plant, such as interactions between phloem-feeders and leaf chewers, as well as interactions between herbivores and pollinators, or herbivores and ovipositing insects.
Case study: aboveground–belowground plant-mediated insect interactions
Since by definition herbivory does not kill the plant, interactions are
temporally and spatially separated insect species are frequent. Abovegound (AG) – belowground (BG) herbivore interactions provide an excellent case study of
plant-mediation involving both spatial and temporal separation. In addition, since most research involving AG–BG interactions is very recent, it provides a
snapshot of the current work in this field. In this section, I will use AG–BG interactions to examine general principles and current understanding of plant-mediated indirect interactions.
Early work on potential aboveground–belowground interactions centered on the idea that BG herbivores could provoke a plant response similar to that of drought stress due to reduction in plant root biomass. The drought stress
response results in increased concentration of AG nitrogen and carbohydrates.
Therefore it was predicted that BG herbivores could have a positive impact on AG herbivores which would benefit from the improved nutritional quality of plant tissue aboveground. Conversely, AG herbivores could negatively impact BG herbivores as reduction in AG plant tissue would reduce allocation to
belowground biomass (Masters 1992, 1993). This hypothesis was tested by Masters and Brown (1992) with a study of the interaction between a root chewer (Phyllopertha horticola) and a leaf miner (Chromatomyia syngenesiae) on
common sow thistle (Sonchus oleraceous). They found that root herbivory
increased pupal weight of the leaf miner while leaf herbivory reduced growth rate of the root herbivore, a contramensal interaction. While they did not find a
difference in total leaf nitrogen, they proposed that the BG herbivore increased leaf quality whereas leaf mining decreased root biomass, a likely cause of the negative part of the interaction.
However, further testing of this “stress response hypothesis” has not been as promising. For instance, Hunt-Joshi and Blossey (2005) tested it using purple loosestrife (Lythrum salicaria) and its specialist herbivores, a leaf-feeding
chrysomelid beetle and a root-feeding weevil. They demonstrated negative effects of leaf herbivory on the weevil in a potted plant study but not under field conditions. They found no effect on the AG beetle in response to the BG weevil in either the potted plant experiment or four-year long field study. The authors suggest that the AG herbivore would be most likely to negatively impact the BG herbivore in cases of extreme defoliation, which would result in partial or
complete death of BG tissue (Hunt-Joshi and Blossey 2005).
Also contrary to the expectations of the stress response hypothesis, Bezemer et al. (2003) found that root herbivory had a negative effect on an aboveground herbivore. They investigated the impact of leaf and root herbivory on terpenoid concentrations in cotton (Gossypium herbaceum). Root herbivory increased terpenoid levels throughout the plant while foliar herbivory increased terpenoids primarily in young leaves. As a result, in the foliar herbivory treatment, the AG herbivore shifted its feeding from young to mature leaves. No response was seen in the BG herbivory treatment. Thus the consumption and growth rate of the AG herbivore was reduced on plants exposed to root herbivory.
In a review of the available literature, Blossey and Hunt-Joshi (2003) suggest that the success of the stress response hypothesis may depend on the study system. The bulk of studies supporting the hypothesis were short term and conducted in early successional communities predominated by annual species.
Blossey and Hunt-Joshi argue that as resource availability and root herbivore populations increase in maturing communities the hypothesis may no longer apply (Blossey and Hunt-Joshi 2003).
The mechanisms governing these patterns may not be adequately predicted by the stress response hypothesis, but some general patterns have begun to emerge. BG organisms can increase or decrease concentrations in AG tissue of putative defense compounds such as terpenoids, glucosinolates and phenolics. For instance, most studies of root-chewing insects show an overall increase in plant chemistry (Bezemer and van Dam 2005, Kaplan et al. 2008).
In contrast to the effect of BG herbivores on AG defenses, the effect of AG herbivores on BG defenses appears to be weak (Soler et al. 2007, Kaplan et al.
2008). There are studies that document either increased or decreased levels of defense chemistry in plant roots as a result of AG herbivory. The quality of root exudates may also be altered by AG herbivory (Bezemer and Van Dam 2005, Soler et al. 2007). These changes may be responsible for changes in
belowground soil communities. For instance, foliar application of JA and SA have been shown to reduce the numbers of feeding grape phylloxera and root-knot nematodes respectively (Bezemer and Van Dam 2005). !
In a recent meta-analysis (Johnson 2012) identified four factors that most influenced experimental results of AG–BG interactions: 1) sequence of herbivore arrival (discussed previously), 2) performance parameter measured, 3) plant and study type, and 4) herbivore type. The performance parameter measured was important when examining effects of AG herbivores on BG herbivores. For
instance, AG herbivory negatively impacted BG survival but positively affected BG population growth rates. In contrast, BG herbivore effects on AG herbivores did not depend on which performance parameters were measured. This finding indicates that researchers should measure multiple performance parameters in order to demonstrate positive or negative interactions between insect
herbivores.
The most consistent results were obtained in lab studies rather than field experiments although it did not matter whether the plant was a natural species or an agricultural plant. Plant type also affected study results with AG herbivores negatively impacting BG herbivores on annuals, but not on perennials. Currently, much of the existing research has been done on annual plant systems. Additional work is needed to understand perennial plant systems.
Herbivore taxa also influenced outcome, perhaps through specificity of elicitation. BG dipterans negatively affected AG herbivores while BG
coleopterans had positive impacts on AG homopterans and negative impacts on AG hymenopterans. AG herbivore type did not appear to have significant effects on BG herbivory. Further research that integrates AG–BG experimental designs with work on specificity of elicitation might clarify this complexity.
Plant-mediated indirect interactions involving natural enemy recruitment!
In addition to mediating interactions between two insect herbivores, plants may mediate more complex interactions involving additional trophic levels.
Several mechanisms by which plants can mediate these interactions have been
identified. Induced plants may alter the herbivore quality or other traits as perceived by predators or parasitoids of herbivores. Plants may also recruit
natural enemies of herbivores, such as parasitoids or predators, thereby reducing herbivore damage or abundance.
While natural enemy attraction is generally considered an indirect defense, only a few studies actually link natural enemy attraction to improved plant fitness (Van Der Meijden and Klinkhamer 2000, Kessler and Baldwin 2002).
Several studies have demonstrated increased seed production on plants with parasitized caterpillars lending credence to the adaptivity of parasitoid attraction (Dicke et. al 2003). However, parasitism of insect herbivores may have some negative impacts on plant fitness. In particular, koinobont parasitoids (those that permit their host to continue development) may slow down herbivore
development, resulting in longer feeding time on the host plant (Dicke 2000).
There is some evidence that hosts parasitized by solitary parasitoid species consume less plant material than unparasitized herbivores. However, hosts parasitized by gregarious parasitoids may consume the same or slightly more than unparasitized hosts (Dicke 2000).
Methods of natural enemy attraction include the release of volatile organic compounds (VOCs) and non-volatile contact cues as well as the secretion of extrafloral nectar (EFN). Upon damage by herbivores, most plants emit VOCs.
There is ample evidence that predators and parasitoids use VOCs for host location (De Moraes et al. 1998, Dicke 2000, Rutledge 1996). Infochemicals (including VOCs and contact cues) may be classified into several types (Dicke
and Sabelis 1988, Rutledge 1996). Kairomones, such as plant VOCs used as natural enemy attractants, benefit both the emitting and receiving organisms. On the other hand, synomones benefit only the receiving organism. For example, parasitoids may use herbivore-derived infochemicals for host location. However, herbivore-based cues are generally more difficult to detect than plant-based cues as there is an evolutionary disadvantage for the herbivore to broadcast its
location (Vet and Dicke 1992). In contrast, plants potentially benefit from
releasing host-location cues (Kessler and Heil 2011). Parasitoid search behavior has been connected to the composition of VOCs as blends are often specific to the initiating herbivore and can be modified by subsequent or simultaneous herbivory.
Natural enemies also use VOCs to discriminate between plants either damaged by previously parasitized or unparasitized hosts. Fatouros (2005) demonstrated that the parasitoid wasps Cotesia rubecula and C. glomerata land preferentially on plants in an unparasitized host treatment rather than the
parasitized host treatment. Plants produced larger quantities of VOCs in the unparasitized host treatment, potentially as an adaptation to reduce induction costs once natural enemies have been recruited (Fatouros 2005).
VOC-mediated enemy attraction has most commonly been studied aboveground, but research of belowground systems demonstrates that root VOCs can also attract natural enemies. In a recent study by Rasmann et al.
(2011), entomopathogenic nematodes were attracted by VOCs released by plant roots damaged by a root-boring beetle. With nematodes present, beetle survival
rate was reduced, leading to no change in aboveground biomass relative to control plants. Without nematodes, the beetles survived and reduced AG
biomass by over 40% (Rasmann et al. 2011). Since AG herbivory can affect BG root exudates and potentially BG VOCs, it is possible that AG herbivory can impact BG indirect defenses. Bezemer and Van Dam (2005) reported that no studies had yet been done to address that question, which could be an important area of future research.
VOCs communicate plant location to insects from a distance. In contrast, contact cues assist parasitoids in detecting a suitable host insect once on the plant. Shiojiri et al. (2001) found that two species of specialist endoparasitic wasps discriminated between plants depending on whether plants had been damaged by host or non-host species. Both wasps spent more time searching on plant leaves damaged by their hosts than on those damaged by nonhost larvae or artificially damaged. Simultaneous damage by their hosts altered both
chemical cues and parasitoid preference. Cotesia glomerata was more attracted to and had higher parasitism rates on plants with simultaneous herbivory than control plants. The responses of Cotesia plutellae to simultaneous herbivory were the opposite (Shiojiri et al. 2001, Shiojiri et al. 2002).
The use of VOCs to attract natural enemies does not come without cost to the plant. In addition to the physiological costs of VOC induction for the plant, there are potential ecological costs as herbivores may also use VOCs to locate the plant (Dicke 2000).
Recent work shows that plants also recruit natural enemies through the secretion of extrafloral nectar (EFN; Kessler and Heil 2011). While studies of this are increasingly common, there is far less research on this type of indirect
defense than on VOC emissions (Heil et al. 2001, Dicke et al. 2003). Extrafloral nectar has been documented in at least 104 plant families (Kessler and Heil 2011) and it has been shown to attract diverse organisms, most commonly ants, wasps, mosquitoes and other insects (Kost and Heil 2005). The induction of EFN through herbivory has been demonstrated in some plants although others
express EFN constitutively (Heil et al. 2001). EFN can indirectly reduce herbivory through the recruitment of ants and wasps that defend the plant from other
insects, including herbivores. For example, Kost and Heil (2005) demonstrated that the application of artificial nectar (simulating natural levels of EFN
production) increased the presence of ants, wasps and flies on lima bean plants (Phaseolus lunatus). This resulted in a significant reduction in the rate of
herbivory on plants treated with artificial nectar relative to control plants.
Surprisingly, induced plant responses may negatively impact natural enemies. For instance, in one study (Thaler 2002), fewer adult parasitoids
emerged from host caterpillars on induced plants than on controls. This decrease was attributed to decreased herbivore quality on induced plants. In the same study, induced tomato plants decreased the abundance of an aphid predator, possibly due to decreased aphid densities. In addition, VOCs occasionally deter rather than attract natural enemies although attraction does seem to be more prevalent (Dicke 2000).